October 2025 AWS outage

20 October 2025

On 20 October 2025, an outage in the US East (N. Virginia) region of Amazon Web Services (AWS), where our US East API runs, caused requests to that region to fail for about an hour and a half. Our API in Singapore, Europe (Frankfurt) and Australia (Sydney), and our web app, were not affected. This was not a security incident: the outage was caused by a fault in AWS’s own systems.

What happened

At 06:48 UTC, a fault in the automation that manages the DNS records of DynamoDB, AWS’s database service, left DynamoDB’s address in the region empty. Two copies of the automation were running at once: one overwrote the current record with an out-of-date one, and the other then deleted the out-of-date record while cleaning up. Applications and AWS’s own services in the region could no longer reach DynamoDB.

AWS restored the record by 09:25 UTC, and DynamoDB had recovered by 09:40 UTC. Knock-on failures in other AWS services in the region, including launching new servers and balancing traffic between them, lasted until 21:20 UTC.

How APITemplate.io was affected

Our monitors, which check each service once a minute, recorded these failures on 20 October 2025 (times in UTC):

• API in US East (N. Virginia): requests failed for a minute at 07:32, then continuously from 07:44 to 09:21, for about 1 hour 38 minutes in total.
• API in Singapore, Europe (Frankfurt) and Australia (Sydney), and the web app (app.apitemplate.io): no failures.

Failed requests received an HTTP 500 error. Our monitors recorded no further failures after 09:21 UTC, although other AWS services in the region did not fully recover until 21:20 UTC.

What you should do

• If you send requests to our US East endpoints (rest-us.apitemplate.io or rest-alt-us.apitemplate.io; see Regions and endpoints) and your integration does not retry failed requests, a document you requested between 07:30 and 09:25 UTC on 20 October 2025 may not have been generated. If you still need it, send the request again.
• We recommend retrying any request that fails with a 5xx error after a short wait, so that a brief outage does not cost you a document.
• For current availability, see Service Status, where you can also subscribe to updates.

What AWS is doing

AWS has published a full account of the outage. It has switched off the automation that manages DynamoDB’s DNS records in every region, and says it will fix the fault and add protections against applying incorrect records before switching it back on. It also says it is:

• limiting how much capacity a single load balancer can take out of service when its health checks fail;
• adding tests of how its system for launching servers recovers after a failure; and
• improving how that system slows down incoming work when it is under heavy load.

Questions

Was any of my data exposed?

No. The outage made our US East API unavailable. It did not give anyone access to your data or to our systems.

Who can I contact?

Email support@apitemplate.io. Please don’t include passwords, keys or other secrets in your message.